Jesse Ivers

Python, Scikit-learn, Tensorflow, LightGBM

• Improved delivery time accuracy ~60% while decreasing total promise times to generate 7 figure lift in GMV.
• Formulated custom objective and multi-layer model with custom architecture for residual modeling

Pytest, Docker, Artifactory, LooperPro, Git, GitHub

• Managed and deployed multi-layer testing of complete ML solution (97% test coverage) in Git-driven CI/CD pipeline reducing deployment failures/outages 100%.
• Designed and documented integrated MLOps approach within Walmart ecosystem to reduce future time-to-deployment across company while maintaining security and resource standards.

• Developed and optimized deep-learning models (PyTorch) to predict tumor recurrence with >0.90 ROC-AUC and PR scores across multiple dense, high-dimensional datasets.
• Designed and built custom implementations of state-of-the-art architectures, including ResNet, Inception, RNN, and VAE.
• Improved model generalization through data augmentation and transformation strategies (Torchvision).
• Accelerated training and inference by 100x using GPU parallelization and high-performance computing (Bash, SLURM).
• Trained two junior engineers in deep learning fundamentals, HPC utilization, and PyTorch.

• Developed novel data processing and visualization techniques (MATLAB: Image Processing Toolbox) for high-variability, multidimensional datasets.
• Extracted quantitative endpoints from probabilistic events using nonlinear curve fitting and Fourier transforms (MATLAB: Curve Fitting Toolbox).
• Implemented statistical filtering techniques to enhance signal quality and remove noise.
• Unsupervised modeling to predict subpopulation abundances using Gaussian Mixture Modelling.
  

• Designed modular, high-throughput simulation framework (Python, NumPy), executing 50,000+ probabilistic simulations to model complex biophysical forward problem.
• Built SQL database (SQLite) to store and query simulated results, improving retrieval efficiency.
• Automated disparate hardware/software components (MATLAB) to reduce data acquisition time by 10x.
• Solved inverse problems from real-world imaging data using multivariate optimization and 3D interpolation (SciPy).
  

• Conducted 100+ industry interviews to refine product-market fit for an emerging imaging technology.
• Led a three-person team in strategic decision-making for potential commercialization.
• Presented weekly project updates to business experts, incorporating feedback into product development.

• Led cross-functional leadership teams to improve organizational strategy and decision-making.
• Managed and trained a cross-cultural team in language education and leadership development.
• Designed and delivered bilingual training programs to enhance communication and soft skills.

• Built pipeline to process and analyze dense, high-dimensional image dataset
• Innovated data analysis and visualization method

Preprint

• A/B tested dozens of deep-learning regression and classification models and optimized hyperparameters in Pytorch
• Developed data processing pipeline for from acquisition to model output
• Accelerated study outputs through remote computing on high-performance cluster and GPU parallelization.
• Quantified and compared performance of array of image data stacks and models
• Summarized and communicated results to collaboration team for grant and publication preparation

Diaz PM, Ivers JD, Quinn KP, Rajaram N. Method for the classification of risk of recurence in early-stage non-small cell lung cancer using deep convolutional residual network modeling in label-free optical images. US Provisional Pat. App. No. 63/862,476, 2025 (Patent-pending)

Ivers JD, Narasimhan R. Platform for Ep-illumination Cross-polarized Hyperspectral Darkfield and Multiphoton Microscopy for Oxygenation and Metabolic Imaging in vivo. US Provisional Pat. App. No. 63/839/813, 2025 (Patent-pending)

Quinn KP, Ivers JD, Rajaram N, Powell N. Method for the classification of the risk of recurrence in invasive breast cancer using deep convolutional neural network modeling in label-free multiphoton microscopy images of formalin-fixed, paraffin-embedded biopsy tissue. (Application under review)

Investigating in vivo tumor biomolecular changes following radiation therapy using Raman spectroscopy

Investigating the relationship between hypoxia, hypoxia-inducible factor 1, and the optical redox ratio in response to radiation therapy

Evaluating differences in optical properties of indolent and aggressive murine breast tumors using quantitative diffuse reflectance spectroscopy

Raman spectroscopy reveals phenotype switches in breast cancer metastasis

Raman Spectroscopy and Machine Learning Reveals Early Tumor Microenvironmental Changes Induced by Immunotherapy

Oral Presentations

Optical metabolic imaging reveals differences in radiation resistant and susceptible tumor xenografts
SPIE Photonics West

Optical metabolic imaging of radiation resistance in head and neck cancer
AIMRC Seminar Series

Poster Presentations

Investigating the relationship between hypoxia, hypoxia-inducible factor 1 (HIF-1), and the optical redox ratio in response to radiation therapy
Winthrop P. Rockefeller Cancer Institute Research Retreat

Resistant Cancer Looks Different
AIMRC 3rd Annual Research Symposium

Multimodal metabolic imaging and proteomics of radiation resistance in head and neck squamous cell carcinoma
Proceedings of the American Association for Cancer Research Annual Meeting 2023

Optical imaging of radiation induced metabolic and molecular changes in radiation sensitive and resistant head and neck cancer
AIMRC 2nd Annual Research Symposium